Journal of Sedimentary Research, 2006, v. 76, 903–918
Research Article
DOI: 10.2110/jsr.2006.078
EOCENE TURBIDITE-POPULATION STATISTICS FROM SHELF EDGE TO BASIN FLOOR,
SPITSBERGEN, SVALBARD
BRYN E. CLARK1
AND
RON J. STEEL2
1
1330 E. 2nd Avenue, #E18T, Anchorage, Alaska 99501, U.S.A.
2
University of Texas at Austin, Geological Sciences, 1 University Station C1100, Austin, Texas 78712, U.S.A.
e-mail: brynclark@hotmail.com
ABSTRACT: There is a need for better tools in the interpretation of depositional sub-environments of deep-marine facies. Using
seismic-scale outcrops of Eocene shelf-margin clinoforms in Spitsbergen, we have statistically characterized the turbidite beds
in terms of grain size, bed thickness, and dominant sedimentary structure at four different lowstand sites in individual,
mappable clinoforms. These sub-environments include: (1) upper-slope canyon and gully fill, (2) upper to middle-slope late
prograding wedge, (3) lower-slope channel–levee complex, and (4) basin-floor fan.
The falling-stage basin-floor fan and slope canyon/gully deposits are generally coarser grained than the rising-stage subenvironments. They also have far fewer siltstone and mudstone beds and have significantly more beds of upper-medium to very
coarse grain size compared to the rising-stage channel–levee and prograding-wedge deposits. Thin beds are particularly
voluminous in the late prograding-wedge and channel–levee systems, whereas only the basin-floor fans and canyon fill have
large numbers of beds thicker than 10 cm.
Turbidity currents deposit sediment at distinctly different sites between shelf edge and basin floor at different times during
the development of the base-level cycle. Differences between these sub-environments are both qualitative and quantitative and
are demonstrable using statistical analysis. The results of this statistical analysis from well-exposed, easily identifiable
architectural elements can be used to reconstruct the shelf margins of analogous basins, where only sparse outcrop or well data
are available.
INTRODUCTION
A series of Eocene shelf-margin to basin-floor clinoforms are exposed
along the mountainsides of Van Keulenfjorden in the Central Tertiary
Basin of Spitsbergen (Fig. 1) (Kellogg 1975; Steel et al. 1985; HellandHansen 1992). These clinoforms contain deep-water sediment-gravityflow deposits in their upper, middle, and lower slope reaches (PlinkBjörklund et al. 2001; Mellere et al. 2002) as well as shallow-water facies
at the shelf–slope break (Steel et al. 2000). The large-scale outcrops on
Spitsbergen (Fig. 2) are very unusual in that they exhibit entire shelfmargin transects exposed in continuous, seismic-scale outcrop. Such
spectacular outcrops make a characterization of the different components
of the lowstand turbidite system particularly powerful because, in order
for such a study to be meaningful, location within the clinoform profile
must be known.
From the series of some 20 clinoforms in Van Keulenfjorden, one with
an obvious, attached deepwater-lowstand complex has been selected
(Clinoform 14 in Fig. 3; Steel and Olsen 2002). Data from this clinoform
are supplemented by data from two other similar clinoforms from the
transect (Clinoforms 8 and 12 in Fig. 3).
In these lowstand complexes, detailed sedimentological sections have
been logged in each of four turbidite sub-environments that have
a particular space–time relationship to each other on the clinoforms. In
order of age, these environmental sites are: (1) upper-slope canyon/gully
fill, (2) basin-floor fan, (3) lower-slope channel–levee complex, and
(4) upper and middle-slope late prograding wedge. These sections were
Copyright E 2006, SEPM (Society for Sedimentary Geology)
logged with the following objectives in mind: (1) to test the hypothesis
that, within the lowstand systems tract, the turbidite beds in each of the
sub-environments should differ significantly from those in the other subenvironments and (2) to attempt to qualify and quantify those differences
using statistical analysis.
In a broader context, a characterization of the turbidite populations in
the different components of the lowstand deepwater sand complex will
better allow the individual lowstand components to be identified when
only local data windows (small, discontinuous outcrops or well data) are
available. This recognition is important for reconstructing the orientation
and size of the system.
THE LOWSTAND SYSTEMS TRACT CONCEPT
Four Components of the Deepwater Lowstand Systems Tract
and Their Timing
The model of Posamentier et al. (1991) suggests that, during the fall
and subsequent rise in relative sea level that produces the lowstand
systems tract, the system components develop in a predictable space–time
manner. Block diagrams illustrating this three-dimensional space–time
relationship between the early and late components of the lowstand
systems tract are shown in Figure 4.
Although there is still disagreement in the literature regarding the
absolute timing relationship of these components in the sea-level curve
(e.g., Kolla and Perlmutter 1993), there is little or no disagreement about
1527-1404/06/076-903/$03.00
904
B.E. CLARK AND R.J. STEEL
JSR
FIG. 1.—General geologic basement map of
the island of Spitsbergen and immediately
surrounding area. After Muller and Spielhagen
(1990). HF 5 Hornsund Fault Zone,
WB 5 Western Boundary Fault,
LF 5 Lomfjorden Fault Zone, NF 5 Ny
Friesland. Dashed box indicates study area.
G 5 Grøndalen, L 5 Liteldalfjellet,
B 5 Brogniartfjellet, S 5 Storvola,
H 5 Hyrnestabben.
the order in which these components occur in the lowstand systems tract.
Examples of Quaternary systems that follow this pattern accurately are
the Amazon Fan (Damuth et al. 1983) and the Nile and Rhone fans
(Bellaiche and Mart 1995).
As sea level falls below the shelf-edge level, gullies and canyons are
incised into the shelf edge and upper slope (Fig. 4A) in response to
lowered base level and steepening of the gradient. The destabilized and
eroded debris is transported by sliding and slumping out onto the slope to
form the mass-transport complex. Subsequent sediment discharge is
funneled through the slope channels to build the basin-floor fan
(Fig. 4A). The sequence boundary is conventionally placed both below
the canyon deposits and below the basin-floor fan.
During the initial stages of sea-level rise, the system begins to backstep,
retreating up the slope so that the finer-grained, heterolithic channel–
levee complex is deposited immediately above, but offset somewhat
upslope of the now-abandoned sand-rich basin-floor fan (Fig. 4B). At
this time, the heads of the canyons are still bypass zones, but the distal
ends have become zones of sedimentation (a result of the rising and
landward-migrating base level). As the fluvial–deltaic system still provides
sediment, the canyons and gullies generally become largely filled, and the
disrupted slope topography is smoothed and healed by the time the sea
level reattains the shelf edge.
Eventually, with continued rise in relative sea level above the shelf edge,
the fluvial–deltaic system reestablishes itself at the shelf margin and,
JSR
TURBIDITE-POPULATION CHARACTERISTICS: A STATISTICAL ANALYSIS
905
FIG. 2.— Helicopter photo of Storvola.
Clinoforms 12 and 14 can be seen clearly in this
photo. A) In photo A, the switch from the
channel–levee complex to the late prograding
wedge on Clinoform 14 can be seen at the point
indicated by the small arrow. The change is
represented by a distinct color change (see
photo A) from the lighter-colored, sandier
channel–levee complex at right to the much
darker, muddier late prograding wedge at left.
Basin-floor fan of Clinoform 14 is out of the
picture to the right. B) Interpretations in photo
B: green 5 upper slope channels, light
brown 5 late prograding wedge, orange 5
channel–levee complex, red 5 basin-floor fan.
provided that sediment discharge is still high, progrades out across the outer
shelf and upper slope again, downlapping the deposits of the older channel–
levee complex (Fig. 4C, D). These deposits can even sometimes reach as far
out as to downlap the proximal fan deposits. During this time the basinfloor-fan deposits are draped mainly by hemipelagic mud (Mitchum 1985;
Vail 1987; Posamentier and Vail 1988; Normark and Piper 1991;
Posamentier et al. 1991; Walker 1992; Normark et al. 1993).
Although it appears as a simple clinoform from a distance, Clinoform
14 has a significant time–space complexity as regards sandstone and
mudstone distribution. Figure 5 illustrates that turbidites are deposited at
three discrete time horizons in this complex, in what can be informally
referred to as early, middle, and late lowstand (labeled A, B, and C,
respectively, in Fig. 5).
Turbidite Sites in the Lowstand Segment of Clinoform 14
CLINOFORMS WITH LOWSTAND SYSTEMS TRACTS IN VAN KEULENFJORDEN
The clinoforms displayed along the Van Keulenfjorden transect
(Fig. 3) have compacted thicknesses of 200–400 m. Clinoform 14 on the
mountain Storvola (Fig. 2) has a decompacted height of about
300 meters, indicating a lowstand water depth of similar magnitude
(height from shelf edge to basin floor). We can interpret sea level to have
fallen to the shelf edge because of the fluvial incision evident in the shelfedge canyons.
The four sub-environments from which turbidite beds have been
sampled are: (1) upper-slope canyon or gully fill, (2) upper-slope late
prograding wedge, (3) lower-slope channel–levee complex, and (4) basinfloor fan. These are easily recognized and differentiated in Clinoform 14
of the Storvola outcrops (Fig. 5). Their location below the shelf–slope
break, regular and predictable time–space arrangement, stratigraphic
relationships, and character all allow them to be designated as
components of a lowstand systems tract.
906
B.E. CLARK AND R.J. STEEL
JSR
FIG. 3.—Interpretive panel showing northern wall of Van Keulenfjorden. The late prograding wedge (brown) and channel–levee complex (orange) sections measured
for this project are from Clinoform 14 on Storvola. Upper-slope channel (orange) sections are also from Clinoform 14, but canyon-fill deposits (dark green) were
measured from Clinoform 9 on Brogniartfjellet. Basin-floor-fan (red) sections were measured from Clinoform 12 on Storvola and Clinoform 14 on Hyrnestabben. All
sections are from the Battfjellet Formation. (Modified from Steel and Olsen 2002.)
FIG. 4.—Block diagram showing space-time relationship between different components of the lowstand systems tract. A) Canyon/gully incision and lowstand (basinfloor) fan deposition, B) early lowstand wedge (channel–levee complex), and C) late lowstand wedge prograding complex (late prograding wedge). D) Axial section in
illustrates stratal geometries and relationships between components. Modified from Posamentier et al. (1991).
JSR
TURBIDITE-POPULATION CHARACTERISTICS: A STATISTICAL ANALYSIS
907
FIG. 5.— Details of sites of turbidite accumulation in time and space in Clinoform 14.
Inset: simplified relative sea-level curve.
TST 5 transgressive systems tract,
HST 5 highstand systems tract,
FSST 5 falling-stage systems tract,
LST 5 lowstand systems tract. Possible timing of events in the LST: (A) falling sea level
causes rivers to incise canyons on shelf edge,
resulting sediment deposited as basin-floor fan;
(B) early rise in sea level causes abandonment
of basin-floor fan, muddier channel–levee
complex is deposited on top as the system
retreats, canyon heads are still zones of bypass,
but distal ends have become zones of sedimentation; and (C) sea level continues to rise,
system retreats further, deltas begin to infill
canyons and prograde across the shelf while the
basin-floor fan is draped by mud.
FIG. 6.— View to north of canyon fill on
Brogniartfjellet (from Clinoform 8, see Fig. 3).
Thickness of sandstone body between dashed
lines increases from , 4.5 m at left to , 16 m
at right.
FIG. 7.— Black arrows indicate upper slope
gullies from Clinoform 12 (see Fig. 3) on the
western (most proximal) shoulder of Storvola.
Discontinuous nature of outcrop reflects laterally discontinuous nature of channelized sand
bodies. Darker, overlying outcrops are proximal late-prograding-wedge deposits. Note
downlapping relationship between these outcrops and underlying gullies (indicated by
white lines).
908
B.E. CLARK AND R.J. STEEL
FIG. 8.— Measured vertical sections through canyon/gully-fill turbidite beds in Clinoform 14 on Brogniartfjellet, Van Keulenfjorden.
JSR
JSR
TURBIDITE-POPULATION CHARACTERISTICS: A STATISTICAL ANALYSIS
909
FIG. 9.— Evidence of scouring seen in
canyon/gully fill (white arrows). Lower arrow
shows scour cutting beds, which exhibit
water-escape structures. Lens cap for scale
(black arrow).
Canyon and Gully Fill.—The canyon and gully-fill deposits are easy to
identify because of their position relatively high on the slope, close to the
shelf–slope break. The term ‘‘canyon’’ refers to the larger-scale (tens of
meters) incisional features, which exhibit obvious large, channeled
geometries (Fig. 6). ‘‘Gullies’’ are smaller-scale (a few meters), more
subtle features and can be recognized by the discontinuous nature of their
outcrops (Fig. 7). Despite the difference in scale between these two
features, they share the same mechanism of generation and position on
the slope, and so are referred to in this paper collectively and
interchangeably under the general term ‘‘canyon/gully.’’
On the outcrop scale, the canyon/gully-fill deposits are composed
mostly of very clean, thick and relatively coarse-grained sandstone beds
that tend to have uneven and erosional lower contacts with the finergrained beds of the underlying deposits. The deposits commonly exhibit
cut-and-fill structures (Figs. 8, 9), evidencing deposition punctuated by
erosion, as could be expected in an environment that is primarily a zone
of bypass. In this sense, the turbidite beds ‘‘collected’’ from the fill of the
gullies and canyons represent a somewhat anomalous turbidite population. Most flows would typically have bypassed the collection sites,
continuing down onto the fans, whereas the beds measured at the
collection sites were anomalously deposited at these sites.
It should also be noted that, given the two-dimensional nature of
outcrop data, these outcrops may not capture the full channel profiles
and so the logged sections may not be perfect representations of the
channel fill.
Basin-Floor Fan.—The time-equivalent basin-floor-fan deposits for
Clinoform 14 are also composed largely of relatively clean, relatively
thick, fine- to medium-grained sandstone beds. However, these sandstone
beds are found at the distal end of the system, from the toe of slope
extending out onto the basin floor (Fig. 5). The change from confined to
unconfined flow and the decrease in slope angle from the slope to the
basin floor caused a decrease in velocity of the canyon-fed flows and
resulted in deposition. The basin-floor system is largely depositional
with very little evidence of deep (. 1 m) scouring in these relatively flatlying beds. A succession of basin-floor turbidite beds is shown in
Figure 10.
Channel–levee Complex.—Channel–levee complexes, which overlie the
basin-floor fan with slopeward offset (Fig. 11) (Posamentier et al. 1991),
can be recognized just above and updip from the basin-floor-fan sands on
the lower part of the slope. The channels show up as discontinuous sandy
outcrops, whereas the levees are represented by more easily eroded,
muddier intervals. Example sections are shown in Figure 12.
Late Prograding Wedge.—The late-prograding-wedge complex, which
represents shelf-edge deltas that prograde and downlap onto the channel–
levee complex after relative sea level regained the shelf edge, is represented
by thick successions of thin bedded, muddy deposits (Figs. 13, 14) that lie
stratigraphically directly above as well as updip from the channel–levee
complexes (Fig. 5). The shift from channel–levee complex to the
prograding wedge is apparent in a distinct color change found about
half way up the slope (Fig. 2) from lighter-colored, sandier deposits
downslope to much darker, muddier deposits upslope. The extremely
muddy nature of these deposits, which rarely contain sandstone beds
thicker than 5–10 cm or coarser than very fine sand, implies a very lowenergy environment just prior to the system transgressing back across the
shelf.
910
JSR
B.E. CLARK AND R.J. STEEL
FIG. 10.— Measured vertical sections through turbidite beds of basin-floor fan in Clinoform 12 on Storvola, Van Keulenfjorden.
Differences and Similarities with Conventional Lowstand Models
Although the Van Keulenfjorden clinoform system is of relatively small
scale (water depth , 500 m) compared to continental-margin clinoforms
(kilometers water depth), there are strong similarities in (1) the presence
of an ordered series of turbidite-bearing, lowstand components (earlystage slope channels, canyons, and fans; later channel–levee systems
positioned on the lower slope) (see Kolla 1993), (2) the slope gradients
below the shelf break of 2–5u (most modern accretionary slopes fall in this
range), and (3) the presence of a late-stage heterolithic, prograding
complex that downlaps the earlier channels and fan (e.g., Posamentier et
al. 1991). However, we do identify a significant difference between the
clinoform architecture in our study and those previously published. Many
of the Van Keulenfjorden clinoforms show a thick (. 10 m) mud-prone
succession in their mid-lowstand segments that develops after fan
abandonment and before the downlapping of the late prograding
complex. This implies significant aggradation of the entire clinoform at
this time and contrasts with conventional lowstand models that show
slope onlap after fan development (Vail et al. 1977; Posamentier et al.
1991).
ANALYSIS OF TURBIDITE BED POPULATIONS
Methods
Several stratigraphic sections were measured in each of the four subenvironments and then, with sections grouped by sub-environment, three
characteristics of each bed were identified in a section: (1) bed thickness,
(2) grain size, and (3) dominant sedimentary structure. Percent mud
content for each section as a whole were also determined.
Bed-thickness values were rounded to the nearest centimeter and
divided into categories defined by a range of thickness values (i.e.,
0–10 cm, 11–20 cm, 21–30 cm, etc.). This grouping was done for
consistency with the other, more qualitative data types. Raw thickness
statistics for the data from Van Keulenfjorden are included in Table 1.
For beds with uneven thickness, measurements were taken from the
thickest part of each bed. For beds with eroded tops, the maximum
observed thickness can be assumed to be the closest possible measurement
to the original thickness because at least that much sediment must have
been present before erosion. For beds with erosional bases, the maximum
thickness best represents the volume of sediment present during
deposition.
JSR
TURBIDITE-POPULATION CHARACTERISTICS: A STATISTICAL ANALYSIS
911
FIG. 11.—A) Photo and B) interpretation of
‘‘backstepping’’ channel complexes: basinward
termination of each channel complex lies
slopeward of that of the underlying (older) one.
Basin-floor fan belongs to an older clinoform
(Clinoform 12, see Fig. 3).
The maximum grain size of each bed was used for determining its
grain-size category. For example, a bed composed of upper medium sand
at its base that fined to upper fine sand at its top would be counted as
upper medium. This was done so that the count reflected the maximum
competence of the current that deposited the bed.
Sedimentary structures were ascertained by using the dominant
sedimentary structure comprising most of the thickness of the bed (see
Tables 2 and 3 for sedimentary-structure categories).
Mud content was defined as the percent of the thickness of sections in
a sub-environment made up of beds with a grain size finer than lower very
fine sand and was determined by dividing the cumulative thickness of all
beds logged in that sub-environment that were composed of silt-size, or
finer, particles by the cumulative thickness of all beds measured in that
sub-environment.
Once all the beds had been counted, the data were normalized by
dividing the number of beds from each category by the total number of
beds counted for that characteristic in all the sections measured in each
sub-environment. This gives a number that represents the percentage of
the total number of beds in a given sub-environment that fall into a given
category. For example, for the characteristic of grain size, 280 different
beds were measured in the canyon/gully-fill sub-environment. Of these
280 beds, 143 fell into the lower-fine-sand category. Therefore
143/280 5 0.51, or 51%, of all beds found in the canyon/gully fill
sub-environment were composed of lower fine sand.
Data
No single category can be used to characterize individual subenvironments. In each characteristic group, one or two categories are
by far the most common in all the sub-environments (e.g., for the
characteristic of bed thickness, the 0–10 cm category represents over 69%
of all measured beds). However, when the frequencies of beds in this
dominant category are considered relative to the frequencies of beds in
other categories, differences between sub-environments become evident.
At the other end of the spectrum, very small sample sizes do not
produce statistically meaningful results because they are not as
912
B.E. CLARK AND R.J. STEEL
JSR
FIG. 12.— Measured vertical sections through turbidite beds of channel–levee deposits in Clinoform 14 on Storvola, Van Keulenfjorden.
representative of the population as a whole. Therefore, categories that
contained fewer than five individuals were considered insufficiently
populated samples and were thrown out.
Grain size.—The most common grain size is lower fine-grained sand,
accounting for about half of all beds (Fig. 15A). It is also the most
common grain size in all sub-environments except for the prograding
wedge, where it is only slightly exceeded by upper very fine-grained
sand.
Despite containing a large majority of beds that fall into the category of
upper fine sand, canyon/gully-fill deposits have a ‘‘tail’’ that extends to
much coarser grain sizes than any of the other sub-environments. While
not quite as coarse as those of the canyon/gully fill, the deposits of the
basin-floor fan show a much larger percentage of grain sizes of upper
fine-grained sand and are coarser than the deposits of the prograding
wedge or the channel–levee complex.
The channel–levee complex and the prograding wedge are both
composed of mainly very fine grain sizes. The channel–levee complex is
composed of more than 84% upper very fine and lower fine sand with
very few beds containing anything coarser. Only one bed in the
prograding wedge succession is coarser than upper fine sand.
Mud Content.—The canyon/gully and basin-floor-fan sub-environments are the cleanest at 5.6% and 0.4%, respectively, whereas the
prograding-wedge and channel–levee complex sub-environments are
much muddier, showing much higher values of 28.0% and 7.8%,
respectively (Fig. 15B).
Sedimentary Structures.—This characteristic is not dominated by one
single category like the others are, although certain categories (i.e.,
massive, flat-laminated, and especially, ripple-laminated) are much more
common than others (Fig. 15C).
In the canyon/gully sections, ripple-laminated beds constitute most of
the deposits (56.2%). Massive (20.76%), flat-laminated (14.18%), and
normally graded massive (9.75%) beds make up a lesser, although still
significant, part of this sub-environment. Also, of the four subenvironments, this one contains the highest percentage of deformed beds
(3.04%).
The prograding-wedge deposits are dominated by beds that are flatlaminated (31.29%) or ripple-laminated (65.41%). These two sedimentarystructure categories are the most common in the channel–levee complex
as well, but in the reverse order (flat-laminated is the most common at
42.76% of total vs. ripple-laminated at 27.92%). Unlike the prograding
wedge, however, there are significant numbers of massive beds in this subenvironment (19.43%).
The basin-floor-fan deposits show the distribution most similar to that
of the channel–levee complex, being the only two sub-environments in
which flat-laminated beds (40.64% in this sub-environment) exceed
ripple-laminated ones. However, ripple-laminated and normally graded
massive beds make up a much smaller proportion (only 10.18% and
2.56%, respectively) of the basin-floor-fan deposits, while the percentage
of massive beds is somewhat higher (42.11%).
Bed Thickness.—The dominant bed thickness category in all subenvironments is 0–10 cm; over 69% of all beds measured fall into this
JSR
TURBIDITE-POPULATION CHARACTERISTICS: A STATISTICAL ANALYSIS
913
Statistics
Raw Thickness Statistics.—Data on raw thickness are quantitative, so
an additional suite of statistics can be run on these numerical data.
Summary statistics (Table 1) show that the average thickness is greatest in
basin-floor fan (0.32 m) followed by the canyon/gully fill (0.14 m),
channel–levee complex (0.11 m), and finally the prograding-wedge
deposits (0.05 m). Minimum thickness values are the same for all subenvironments, 0.01 m. Maximum thickness values show a trend similar to
average thickness, but with the canyon/gully fill (4.9 m) and basin-floor
fan (3.0 m) reversed in rank. The thickest bed in the channel–levee
complex deposits was 2.12 m, 1.24 m for the prograding wedge. Standard
deviations follow the same trend as average thickness: basin-floor fan
(0.47), canyon/gully fill (0.42), channel–levee complex (0.22), and
prograding wedge (0.10).
Frequency distributions of the data, which demonstrate how commonly measurements exist in the data distribution, provide a method for
picking out differences that might otherwise be masked by the abundance
of one bed type. For the bed-thickness data, frequency distributions prove
to be a useful tool, highlighting observed trends. Thin beds, such as 2 cm,
have much higher cumulative percentages in the prograding wedge (59%)
and channel–levee complex (59%) than they do in the canyon/gully fill
(13%) or the basin-floor fan (21%).
FIG. 13.—Outcrop photo of late-prograding-wedge deposits on Storvola. Note
relatively thin, muddy nature of beds.
category (Fig. 15D). If only the canyon/gully fill and the prograding
wedge are considered, this percentage jumps to nearly 86% and 92%,
respectively. However, the basin-floor-fan sub-environment has a much
more even distribution of bed-thickness values. In the basin-floor-fan
deposits, beds thicker than 10 cm make up more than half of the bed
population.
Interpretations
The most obvious trend in the data is the marked dissimilarity between
the deposits formed during the sea-level-fall half cycle and those formed
during the rise half cycle. This trend fits well with sequence stratigraphic
predictions (e.g., Posamentier et al. 1991).
Deposition in the canyon/gully sub-environment probably occurs when
relative sea level is at its lowest point; at that time this is the most
proximal sub-environment, so it follows that it should contain the
coarsest material. At the same time, the basin-floor fan is deposited at the
bottom of a steep ‘‘bypass’’ slope, so large volumes of sand are able to
reach this sub-environment, while fines are carried even farther into the
basin.
As relative sea level begins to rise, the system begins to retreat and the
more proximal locations, where coarser materials are trapped, begin to
backstep away from the basin center. Consequently, the progradingwedge and channel–levee complex sub-environments are much thinner
bedded and finer grained.
Analysis of Variance (ANOVA).—To test the hypothesis that there was
a statistically significant difference among the sub-environments, an
analysis of variance (ANOVA) test (described by Scheffler 1980) was
applied to the data. This test compares the variation between values
within a group of data (this is referred to as the ‘‘within-group variance’’
or ‘‘error variance’’ and is assumed to be due to the random variation
present in any data population) with the variation between different
groups (which is called the ‘‘among-group variance’’ and may be due to
outside forcing). If the among-group variance is greater than the withingroup variance, it can be said that the difference between the groups is
statistically significant.
Note that an ANOVA will also pick out differences among the mean
values of the different groups because this contributes to the betweengroups variance.
Turbidite Sub-Environment ANOVA.—ANOVAs were performed to
evaluate variation between the four turbidite sub-environments as
exhibited in each of three categories: grain size, sedimentary structure,
and bed thickness. Because of the construction of the test itself, a separate
ANOVA must be performed for each division in each category (i.e., one
ANOVA to test the hypothesis that there is a significant difference in the
frequency of lower very fine beds between the different sub-environments,
another test to compare upper very fine beds, another for lower fine, etc.).
The resultant F values are given in Table 2 along with critical F values for
comparison.
This table shows a statistically significant difference between treatments, and so the hypothesis is accepted at the 99% confidence level.
Three of the four experiments that produced random results at the 99%
confidence level were rejected because of the fact that all or nearly all of
the individuals in these populations are either zeros or very small numbers
(, 5 individuals), which decreases the significance of the experiment. The
fourth experiment produced nonrandom results at the 95% confidence
level.
Raw Thickness ANOVA.—An ANOVA performed on the unbinned or
raw thickness values to test the hypothesis that there is a significant
difference in raw thickness values between the different sub-environments
gives an F value of 25.72. This is well above the critical F value for the
914
B.E. CLARK AND R.J. STEEL
JSR
FIG. 14.— Measured vertical sections through prograding wedge deposits in Clinoform 14 on Storvola, Van Keulenfjorden.
99% confidence level (4.62), and so the hypothesis can be accepted and it
can be said that there is a significant difference between the raw thickness
values.
Within-Group Variance for Basin-Floor Fan Data.—An ANOVA can
be performed only when values of n (the number of data points in each
data population) for different treatments are of a similar magnitude.
However, there are 47 measured sections from the basin-floor fan
(n 5 47) whereas values of n for the other three sub-environments
ranged only from n 5 7 to n 5 10. Much larger values of n would
have been needed for each of the other three sub-environment
treatments to be able to compare them to the basin-floor-fan treatment.
Therefore, the basin-floor-fan data were split into different populations
depending upon the locality from which the sections were measured and
JSR
TURBIDITE-POPULATION CHARACTERISTICS: A STATISTICAL ANALYSIS
915
TABLE 1.— Summary statistics and ANOVA results for unbinned (raw) thickness data (units are in meters).
Basin-floor Fan
Canyon/Gully Fill
Channel-Levee Complex
Prograding Wedge
ANOVA
272
0.01
3
0.32
0.47
460
0.01
4.9
0.14
0.42
460
0.01
2.13
0.1
0.22
956
0.01
1.24
0.05
0.1
F[3,1000] 5 26.74
Critical F[3,1000].05 5 3.00
Critical F[3,1000].01 5 4.62
n
minimum
maximum
average
S
n 5 number of beds measured.
S 5 standard deviation.
TABLE 2.— Analysis of variance (ANOVA) results for different sub-environments (n 5 31 ).
A. Bed thickness
F value
# of beds
measured
0–10 cm
11–20 cm
21–30 cm
31–40 cm
41–50 cm
51–100 cm
101–150 cm
151–200 cm
. 200 cm
195.13
29.26
37.25
13.22
13.48
21.72
9.26
3.15
5.18
1873
170
87
45
31
69
16
11
10
Critical F values: F[3, 27].05 5 2.96; F[3,
27].01
B. Grain size
F value
# of beds
measured
lower v. fine
upper v. fine
lower fine
upper fine
lower med.
upper med.
lower coarse
upper coarse
lower v. coarse
upper v. coarse
9.3
34.49
187.52
66.82
8.75
4.19
8.7
6.28
0
2.89
98
603
1104
448
44
25
5
4
0
2
C. Dominant sedimentary structure
F value
# of beds
measured
massive
massive-normally graded
massive-reverse graded
massive-reverse to normal
flat laminated
rippled or ripple laminated
climbing ripples
deformed
63.42
11.04
6.14
4.4
77.99
75.08
1.35
31.81
271
70
12
12
542
913
1
47
5 4.60.
TABLE 3.— Analysis of variance (ANOVA) results for different basin floor-fan-locality data (n 5 46).
A. Bed thickness
F value
# of beds
measured
0–10 cm
11–20 cm
21–30 cm
31–40 cm
41–50 cm
51–100 cm
101–150 cm
151–200 cm
. 200 cm
0.35
1.16
0.14
0.69
0.25
0.38
0.63
0.45
1.11
618
257
135
64
53
154
44
28
24
Critical F values: F[4, 41].05 5 2.61; F[4,
41].01
B. Grain size
F value
# of beds
measured
lower v. fine
upper v. fine
lower fine
upper fine
lower med.
upper med.
lower coarse
upper coarse
lower v. coarse
upper v. coarse
0.55
0.34
0.94
0.28
1.66
1.88
0.48
0
0
0
46
120
605
423
103
43
4
0
0
0
C. Dominant sedimentary structure
F value
# of beds
measured
massive
massive-normally graded
massive-reverse graded
massive-reverse to normal
flat laminated
rippled or ripple laminated
climbing ripples
deformed
0.51
2.89
1.02
1.41
0.89
1.01
1.64
0.34
537
33
26
17
523
131
1
14
5 3.83.
an ANOVA was performed on these populations. This ANOVA was
applied to test the hypothesis that there was no statistically significant
difference among basin-floor-fan data collected from different localities.
Results from this ANOVA are shown in Table 3. From this table, one
can see that there is no statistically significant difference between different
basin-floor-fan localities in 26 out of 27 cases. The variance among basinfloor-fan data can be considered random. Basin-floor-fan deposits are
shown to have no statistically significant variability between localities.
The hypothesis is therefore accepted.
t-Test Data.—The initial ANOVAs pointed to a difference somewhere
in the data. As stated above, this difference appeared to be between the
turbidite beds deposited during falling relative sea level (canyon/gully-fill
and basin-floor-fan deposits) and those deposited during its early rise
(channel–levee and prograding-wedge deposits).
Further statistical testing was needed to attempt to tease out where that
difference was occurring, and so t-tests were applied to the grouped data.
A t-test is a more powerful variance test than an ANOVA, but can be
used only to test for differences between two data sets whereas an
ANOVA is used to test differences between three or more groups. The
hypothesis being tested by the t-test was as follows: there is a statistically
significant difference between turbidite beds deposited during falling and
rising relative sea level. Results from this suite of tests are shown in
Table 4.
The t-test results are not quite as sweeping as the ANOVA results, but
there are some important trends to be noted. The first is that the
hypothesis can be accepted at the 99% confidence level in all but four of
the cases in which there were enough data points for the test to be
accepted. For one of these cases it can be accepted at the 95% confidence
level. The second is that the extreme abundance of finer-grained
sediments in all sub-environments seems to mask differences in grainsize distributions.
Significance of the Characterization of Turbidite Bed Populations
A knowledge of the expected range and relative differences in grain
size, bed thickness, and mud content for the two early-stage and two late-
916
B.E. CLARK AND R.J. STEEL
JSR
JSR
TURBIDITE-POPULATION CHARACTERISTICS: A STATISTICAL ANALYSIS
917
TABLE 4.— t-test results for beds deposited during rising vs. falling relative sea level (n 5 31).
A. Bed thickness
t value
# of beds
measured
0–10 cm
11–20 cm
21–30 cm
31–40 cm
41–50 cm
51–100 cm
101–150 cm
151–200 cm
. 200 cm
6.33
3.4
17.11
41.43
19.65
36.78
102.52
70.97
41.26
1873
170
87
45
31
69
16
11
10
B. Grain size
t value
# of beds
measured
lower v. fine
upper v. fine
lower fine
upper fine
lower med.
upper med.
lower coarse
upper coarse
lower v. coarse
upper v. coarse
6.89
6.05
2.14
1.54
1.9
28.39
32.86
47.9
0
40.17
98
603
1104
448
44
25
5
4
0
2
C. Dominant sedimentary structure
t value
# of beds
measured
massive
massive-normally graded
massive-reverse graded
massive-reverse to normal
flat laminated
rippled or ripple laminated
climbing ripples
deformed
12.41
9.55
46.65
28.44
4.28
2.51
204
12.41
271
70
12
12
542
913
1
47
Critical t values: t[29].05 5 1.699; t[29].01 5 2.46.
stage components of the lowstand turbidite complexes attached to such
shelf margins can be important for both academic and applied reasons.
Quantitative data of this kind, reflecting generation and accumulation of
beds during a period of relative sea-level fall and rise, can be used in
evaluation of source-to-sink sediment volume partitioning. The same
knowledge of expected bed population characteristics might allow the
individual turbidite sites to be recognized and oriented where only small
data windows are available, for example in areas of poor outcrop or in
well logs from exploration drilling.
DISCUSSION
Using bed types or perceived trends of beds in the interpretation of
turbidite successions to try to better understand deepwater subenvironments has long been practiced. In the pre–sequence stratigraphy
deepwater world, bed trends of upward coarsening and thickening or
upward thinning and fining were commonly used to recognize submarinefan lobes and channels, respectively (Mutti and Ricci Lucchi 1972;
Walker 1978). Some researchers have claimed that this recognition
method was oversimplified and advocated a more rigorous, clustering
statistical approach (Chen and Hiscott 1999). There is an extensive pool
of literature on bed-scale turbidites, and analysts have emphasized field
description and classification (e.g., Ricci Lucchi and Valmori 1980;
Ghibaudo 1992), statistical modeling (Felletti 2004), reservoir modeling
(Hurst et al. 2000; Slatt 2000) or architectural-element approaches
(Pickering et al. 1995). Most of this bed-scale research has focused on the
turbidites that build submarine fans, and the objective has often been to
work back from turbidite characterization to depositional setting or subenvironment. The research approach presented here has rather been the
opposite. Because we can identify the range of turbidite-population sites
within the seismic-scale, shelf-margin clinoforms, we know both the subenvironments and their temporal development during the shelf-margin
accretion cycle. Our objective has been to characterize the turbidite
populations from known sub-environments. Description and characterization of the turbidites from the late prograding complex is done here,
possibly for the first time. Placement and characterization of the turbidite
populations within the context of the evolving lowstand systems tract is
also new.
CONCLUSIONS
Thin, fine-grained beds dominate all successions. Nevertheless,
differences are detectable, particularly at the other ends of the grainsize and bed-thickness spectrums. Basin-floor-fan and slope canyon/gully
deposits are generally coarser grained. They have far fewer siltstone and
mudstone beds than deposits found in the channel–levee and prograding
wedge as well as significantly more upper-medium to very coarse-grained
beds. Thin beds are particularly voluminous in the late prograding wedge
and channel–levee systems, whereas only the basin-floor fans and canyon
fill have large numbers of beds thicker than 10 cm. These differences can
be expressed quantitatively using summary statistics and analysis-ofvariance tests.
Maximum and average bed-thickness values are higher for the canyon/
gully fill and basin-floor fan than for the channel–levee complex and
prograding-wedge sections (see Table 1). Standard deviations in these
data are also higher for the basin-floor fan and canyon/gully fill because
these sub-environments contain a wider range of bed thickness values
than the almost exclusively thin-bedded channel–levee complex and
prograding wedge.
ANOVA tests of variance show significant differences at the 95%
confidence level for nearly every category. When beds from different
localities in the basin-floor-fan sub-environment are compared using an
ANOVA, no significant difference is evident. This suggests that there is
minimal variability between different basin-floor-fan deposits. Therefore,
these statistical tests can potentially be applied to a wide range of basinfloor fan deposits with similar results. It should be noted that there may
be significant differences between the edges and the more central parts of
the fans but this hypothesis is outside the scope of this study.
t-tests on grouped rising-relative-sea-level vs. falling-relative-sea-level
deposits indicate a significant difference between these two groups,
confirming the quantitative trends described above. However, an extreme
abundance of beds in the lower to upper fine-grained range in all subenvironments tends to mask differences at those grain sizes. Summary
statistics and a frequency distribution on raw (unbinned) thickness data
also confirm this trend.
This characterization of the turbidite bed populations within the
different components of the lowstand complex will better allow these
individual components to be identified when only local data windows,
r
FIG. 15.—Frequency histograms. A) Histogram comparing grain-size frequency in the four sub-environments. B) Comparison of mud content as percents of total for
the different sub-environments. C) Histogram comparing dominant sedimentary-structure frequency in the four sub-environments. D) Histogram comparing bedthickness frequencies in the four sub-environments.
918
B.E. CLARK AND R.J. STEEL
such as well data or small, discontinuous outcrops, are available. This can
be critical to the successful reconstruction of the orientation and size of
the larger system to which the components belong.
ACKNOWLEDGMENTS
We thank the WOLF sponsors BP, BHP, ConocoPhillips, ExxonMobil,
PDVSA, Shell, and Statoil for support and enthusiastic discussions during
this project. We would also like to thank Dag Nummedal, Gregg Cawley,
Paul Heller, Paul Myrow, Victor Pusca, Dave Jennette, and Dave Mohrig for
discussion and helpful technical and stylistic input.
Thanks also to Piret Plink-Björklund, Jeff Crabaugh, Anna Pontén, Louise
Sjögren, Donatella Mellere, Louis Sass, the crews of Johnathan, Farm, Lance,
and Hillerø, and the people at Airlift Helicopters for their indispensable aid in
the field, be it geological or logistical.
REFERENCES
BELLAICHE, G., AND MART, Y., 1995, Morphostructure, growth patterns, and tectonic
control of the Rhone and Nile deep-sea fans: a comparison: American Association of
Petroleum Geologists, Bulletin, v. 79, p. 259–284.
CHEN, C., AND HISCOTT, R.N., 1999, Statistical analysis of facies clustering in submarinefan turbidite successions: Journal of Sedimentary Research, v. 69, p. 505–517.
DAMUTH, J.E., KOWSMANN, R.O., FLOOD, R.D., BELDERSON, R.H., AND GORINI, M.A.,
1983, Age relationships of distributary channels on Amazon deep-sea fan:
implications for fan growth pattern: Geology, v. 11, p. 470–473.
FELLETTI, F., 2004, Statistical modeling and validation of correlation in turbidites: an
example from the Tertiary Piedmont Basin (Castagnola Fm., N. Italy): Marine and
Petroleum Geology, v. 21, p. 23–39.
GHIBAUDO, G., 1992, Subaqueous sediment gravity flow deposits: practical criteria for
their field description and classification: Sedimentology, v. 39, p. 423–454.
HELLAND-HANSEN, W., 1992, Geometry and facies of Tertiary clinothems, Spitsbergen:
Sedimentology, v. 39, p. 1013–1029.
HURST, A., CRONIN, B., AND HARTLEY, A., 2000, Reservoir modeling of sand-rich
deepwater clastics: the necessity of downscaling: Petroleum Geoscience, v. 6,
p. 67–76.
KELLOGG, H.E., 1975, Tertiary stratigraphy and tectonism in Svalbard and continental
drift: American Association of Petroleum Geologists, Bulletin, v. 59, p. 465–485.
KOLLA, V., 1993, Lowstand deep-water siliciclastic depositional systems: characteristics
and terminologies in sequence stratigraphy and sedimentology: Centres de Recherche
Exploration-Production Elf Aquitaine, Bulletin, v. 17, p. 67–78.
KOLLA, V., AND PERLMUTTER, M.A., 1993, Timing of turbidite sedimentation on the
Mississippi Fan: American Association of Petroleum Geologists, Bulletin, v. 77, p.
1129–1141.
MELLERE, D., PLINK-BJÖRKLUND, P., AND STEEL, R., 2002, Anatomy of shelf-edge deltas
on a prograding Eocene shelf margin, Spitsbergen: Sedimentology, v. 49, p.
1181–1206.
MITCHUM, R.M., 1985, Seismic stratigraphic expression of submarine fans, in Berg,
O.R., and Woolverton, D.G., eds., Seismic Stratigraphy II: An Integrated Approach
to Hydrocarbon Exploration: American Association of Petroleum Geologists,
Memoir 39, p. 117–136.
MULLER, R.D., AND SPIELHAGEN, R.F., 1990, Evolution of the Central Tertiary Basin of
Spitsbergen: towards a synthesis of sediment and plate tectonic history: Palaeogeography, Palaeoclimatology, Palaeoecology, v. 80, p. 53–172.
MUTTI, E., AND RICCI LUCCHI, F., 1972, Le torbidite dell’Apennino Setlentrionale:
introduzione all’analisi di facies: Società Geologica Italiana, Memorie, v. 11, p.
161–199.
JSR
NORMARK, W.R., AND PIPER, D.J.W., 1991, Initiation processes and flow evolution of
turbidity currents: implications for the depositional record, in Osborne, R.H., ed.,
From Shoreline to Abyss: SEPM, Special Publication 46, p. 207–230.
NORMARK, W.R., POSAMENTIER, H., AND MUTTI, E., 1993, Turbidite systems: state of the
art and future: Reviews of Geophysics, v. 31, p. 91–116.
PICKERING, K., CLARK, J.D., SMITH, R.D.A., HISCOTT, R.N., RICCI LUCCHI, F., AND
KENYON, N.H., 1995, Architectural element analysis of turbidite systems and selected
topical problems for sand-prone deepwater systems, in Pickering, K.T., Hiscott, R.N.,
Kenyon, N.H., Ricci Lucchi, F., and Smith, R.D.A., eds., Atlas of Deepwater
Environments: Architectural Styles in Deepwater Turbidite Systems: London,
Chapman & Hall, p. 1–10.
PLINK-BJÖRKLUND, P., MELLERE, D., AND STEEL, R., 2001, Turbidite variability and
architecture of sand-prone, deep-water slopes: Eocene clinoforms of the Central
Basin, Spitsbergen: Journal of Sedimentary Research, v. 71, p. 895–912.
POSAMENTIER, H.W., AND VAIL, P.R., 1988, Eustatic controls on clastic deposition II—
sequence and systems tract models, in Wilgus, C.K., Hastings, B.S., Ross, C.A.,
Posamentier, H.W., Van Wagoner, J.C., and Kendall, C.G.S.C., eds., Sea Level
Changes, an Integrated Approach: SEPM, Special Publication 42, p. 125–154.
POSAMENTIER, H.W., ERSKINE, R.D., AND MITCHUM, R.M., JR., 1991, Models for
submarine-fan deposition within a sequence-stratigraphic framework, in Weimer, P.,
and Link, M.H., eds., Seismic Facies and Sedimentary Processes of Submarine Fans
and Related Turbidite Systems: New York, Springer-Verlag, p. 127–136.
RICCI LUCCHI, F., AND VALMORI, E., 1980, Basin-wide turbidites in a Miocene, oversupplied deep-sea plain: a geometrical analysis: Sedimentology, v. 27, p. 241–270.
SCHEFFLER, W.C., 1980, Statistics for the Biological Sciences: Reading, Massachusetts,
Addison-Wesley Publishing, 230 p.
SLATT, R., 2000, Why outcrop characterization of turbidite systems, in Bouma, A., and
Stone, C., eds., Fine-Grained Turbidite Systems: American Association of Petroleum
Geologists, Memoir 72, p. 187–194.
STEEL, R.J., AND OLSEN, T., 2002, Clinoforms, clinoform trajectories and deep-water
sands, in Armentrout, J.M., and Rosen, N.C., eds., Sequence Stratigraphic Models for
Exploration and Production: Evolving Methodology, Emerging Models and
Application Histories: SEPM, Gulf Coast Section, 22nd Annual Bob F. Perkins
Research Conference, p. 367–380.
STEEL, R.J., GJELBERG, J., HELLAND-HANSEN, W., KLEINSPEHN, K., NØTTVEDT, A., AND
RYE-LARSEN, M., 1985, The Tertiary strike-slip basins and orogenic belt of
Spitsbergen, in Biddle, K.T., and Christie-Blick, N., eds., Strike-Slip Deformation,
Basin Formation, and Sedimentation: SEPM, Special Publication 37, p. 339–359.
STEEL, R.J., CRABAUGH, J., SCHELLPEPER, M., MELLERE, D., PLINK-BJÖRKLUND, P.,
DEIBERT, J., AND LOESETH, T., 2000, Deltas vs. rivers on the shelf edge: Their relative
contributions to the growth of shelf-margins and basin-floor fans (Barremian and
Eocene, Spitsbergen), in Weimer, P., Slatt, R.M., Coleman, J., Rosen, N.C., Nelson,
H., Bouma, A.H., Styzen, M.J., and Lawrence, D.T., eds., Deep-Water Reservoirs of
the World: SEPM, Gulf Coast Section, 20th Annual Bob F. Perkins Research
Conference, p. 981–1009.
VAIL, P.R., 1987, Seismic stratigraphy interpretation using sequence stratigraphy part I:
seismic stratigraphy interpretation procedure, in Bally, A.W., ed., Atlas of Seismic
Stratigraphy: American Association of Petroleum Geologists, Studies in Geology 27,
p. 1–10.
VAIL, P.R., MITCHUM, R.M., JR., AND THOMPSON, S., III., 1977, Seismic stratigraphy and
global changes of sea level, Part 3: Relative changes of sea level from coastal onlap, in
Payton, C.E., ed., Seismic Stratigraphy—Applications to Hydrocarbon Exploration:
American Association of Petroleum Geologists, Memoir 26, p. 63–81.
WALKER, R.G., 1978, Deepwater sandstone facies and ancient submarine fans: models
for exploration for stratigraphic traps: American Association of Petroleum
Geologists, Bulletin, v. 62, p. 932–966.
WALKER, R., 1992, Turbidites and submarine fans, in Walker, R., and James, N.P., eds.,
Facies Models: Response to Sea Level Change: Geological Association of Canada, p.
239–263.
Received 11 March 2005; accepted 15 December 2005.